Poglitazone promotes insulin-induced activation of phosphoinositide 3-kinase in 3T3-L1 adipocytes by inhibiting a negative control mechanism

~ Molecularand Cellular Endocrinology

ELSEVIER

Molecular and Cellular Endocrinology103 (1994) 1-12

Pioglitazone promotes insulin-induced activation of phosphoinositide 3kinase in 3T3-L1 adipocytes by inhibiting a negative control mechanism Karen M. Sizer, Craig L. Smith, Cynthia S. Jacob, Michael L. Swanson, John E. Bleasdale* Metabolic Diseases Research, Upjohn Laboratories, Kalamazoo, M149001, USA

Received 3 December 1993; accepted I March 1994

Abstract

Activation of phosphoinositide 3-kinase (PI 3-kinase) is an early event in insulin signal transduction that is blocked completely in adipocytes from insulin-resistant KKAY mice. Treatment of KKAY mice with pioglitazone, an anti-diabetic thiazolidinedione, partially restores insulin-dependent changes in PI 3-kinase. The mechanism of this effect of pioglitazone was investigated using murine 3T3-L1 cells as an experimental model. Insulin and insulin-like growth factor I (IGF-I) each elicited rapid (within 2 min) and large (2-5-fold) increases in PI 3-kinase activity that could be immunoprecipitated using anti-phosphotyrosine (pY) antibodies. Maximal insulininduced activity of PI 3-kinase in pY-immunoprecipitates was similar in 3T3-L1 adipocytes and mouse adipocytes, but the kinetics of activation differed. Insulin- and IGF-I-induced changes in PI 3-kinase were each half-maximal at 3-5 nM of hormone and were not additive. Increases in both insulin-induced and IGF-I-induced pY-immunoprecipitable PI 3-kinase activity were observed when 3T3-LI fibroblasts became confluent and when they adopted the adipocyte phenotype. Pioglitazone (10#M), administered either acutely or chronically to either 3T3-L1 adipocytes or 3T3-L1 fibroblasts, did not greatly alter the kinetics, magnitude or sensitivity of changes in PI 3-kinase elicited by either insulin or IGF-I. In contrast, the attenuation by isoproterenol of insulin-induced changes in PI 3-kinase was prevented in cells pretreated with pioglitazone. This effect of pioglitazone did not involve inhibition of isoproterenol-elicited accumulation of cyclic AMP. Pioglitazone also prevented attenuation of insulin induced changes in PI 3-kinase by cell penetrating analogs of cyclic AMP. Pioglitazone, therefore, has no direct effect on insulin-stimulated PI 3-kinase activity, but interferes with a cyclic AMP-dependent mechanism that normally antagonizes this action of insulin. These data support the proposition that the facilitation of insulin action by pioglitazone involves, at least in part, an inhibition of a negative control mechanism. Keywords: Phosphoinositide 3-kinase; Insulin; Insulin-like growth factor-1; 3T3-L1 cells; Pioglitazone

1. I n t r o d u c t i o n

The binding of insulin to specific cell surface receptors results in a variety of cellular responses that differ kinetically and mechanistically. Typically, however, all these responses are attenuated in the insulin resistance that is associated with non-insulin dependent diabetes mellitus (NIDDM) (Olefsky et al., 1988). This suggests that this insulin resistance involves an early process in signal transduction following the binding of insulin to its receptor which, in most cases, appears to be structurally normal (Olefsky et al., 1988). Furthermore, the insulin resistance of NIDDM (and experimental models of this disease) is ameliorated by the anti-diabetic thiazolidinediones, such as pioglitazone, which appear to sensitize cells to insulin by directly or indirectly facilitating an early event in in* Corresponding author.

sulin action (Colca and Morton, 1990). The mechanism of action of the thiazolidinediones, however, remains unknown. Knowledge of the mechanism of thiazolidinedione action would not only permit mechanism-based searches for new anti-diabetic drugs, but also may help elucidate the molecular defects of insulin resistance. A rapid consequence of the interaction of insulin with the extracellular a-subunits of insulin receptors is the activation of a tyrosine kinase domain in the cytosolic region of the fl-subunits. A major substrate for the insulin receptor tyrosine kinase is a 185 kDa protein (insulin receptor sustrate 1, IRS-1) that has been purified and characterized, and the gene encoding this protein has been sequenced (reviewed by Myers and White, 1993). IRS-1 is rapidly phosphorylated at multiple tyrosine residues by the insulin receptor and becomes a binding target for several proteins that contain one or more regions of homology with a domain first identified in a src oncoprotein

0167-8140/94/$07.00 © 1994 Elsevier Science Ireland Ltd. All rights reserved SSDI 0167-8140(93)03303-G

2

K.M. Sizer et al./Mol. Cell. Endocrinol. 103 (1994) 1-12

eral proteins that contain one or more regions of homology with a domain first identified in a src oncoprotein (SH2 domain) (Sun et al., 1992). The regulatory subunit of phosphoinositide 3-kinase (PI 3-kinase) contains two SH2 domains and this enzyme binds to phosphorylated IRS-1 (Backer et al., 1992; Myers et al., 1992). PI 3kinase is activated either as a direct consequence of this association (Folli et al., 1992), or as a result of subsequent phosphorylation of the regulatory subunit of PI 3-kinase (Hayashi et al., 1992). PI 3-kinase products accumulate rapidly and transiently following activation of insulin receptors (Ruderman et al., 1990; Endemann et al., 1990). The function of PI 3-kinase products (or their metabolites) in insulin signal transduction is unknown, but it has been proposed that phosphatidylinositol 3,4,5-trisphosphate is a specific activator of the ~-isozyme of protein kinase C (Nakanishi et al., 1993). Regardless of the precise function of PI 3-kinase in insulin action, activation of this enzyme is an early event in signal transduction and, as such, is a process that is potentially influenced by thiazolidinediones. Indeed, it was found that insulin elicited a large increase in antiphosphotyrosine immunoprecipitable PI 3-kinase activity in epididymal fat cells isolated from normal C57BL/6J mice and from pioglitazone-fed insulin-resitant KKAy mice, but not in fat cells from untreated KKAY mice (Dailey et al., 1993). The mechanism of this effect of pioglitazone on insulin-stimulated PI 3-kinase activity has been investigated further using 3T3-L1 adipocytes as an experimental model. 3T3-L1 adipocytes respond not only to insulin but also to pioglitazone and other thiazolidinediones (Reed et al., 1977; Kletzien et al., 1992; Stevenson et al., 1990). The purpose of this investigation was to characterize insulin-dependent changes in PI 3-kinase activity in 3T3-L1 cells and to define the effects of piogtitazone on these responses. The data support the conclusion that insulin-dependent changes in PI 3kinase activity increase as 3T3-L1 cells adopt the adipocyte phenotype. Pioglitazone has a negligible direct effect on insulin-induced changes in PI 3-kinase, but may indirectly facilitate this action of insulin by interfering with a mechanism that normally antagonizes insulin-induced changes in PI 3-kinase. 2. Materials and methods

Male C57BL/6J mice (8-12 weeks of age) were from Charles River Laboratories, Wilmington, MA, and were fed Purina mouse chow 5015 and water ad libitum. 3T3L1 cells were obtained from the American Type Culture Collection. Antibodies against phosphotyrosine (affinitypurified rabbit immunoglobulins) were from Zymed Laboratories, South San Francisco, CA, and antibodies against rat PI 3-kinase (rabbit whole antiserum), were obtained from Upstate Biotechnology Inc., Lake Placid, NY. Protein A immobilized on agarose beads was obtained from Boehringer Mannheim Corp., Indianapolis,

IN. [y-32p]ATP (3000 Ci/mmol) and [2,8-3H]adenosine 3',5' cyclic phosphate (45 Ci/mmol) were obtained from DuPont NEN Research Products, Boston, MA. Pioglitazone (5- {4-[2-(5-ethyl-2-pyridinyl)ethoxy]benzyl}thiazolidine-2,4-dione) was obtained from Takeda Chemical Industries Ltd., Osaka, Japan. All other reagents were from Sigma Chemical Co., St. Louis, MO. 2.1. Cell culture and adipocyte conversion

3T3-L1 fibroblasts were maintained and converted to the adipocyte phenotype essentially as described by Green and Kehinde (1975). 3T3-L1 fibroblasts were grown to confluence in 35-ram culture dishes that contained Dulbecco's modified Eagle's high glucose medium (DMEM, Gibco-BRL Inc., Grand Island, NY) supplemented with fetal bovine serum (5%, v/v) and gentamicin (10/zg/ml) (DMEM/5% FBS), in a 10% CO 2 atmosphere at 37°C. Medium was replaced every 2 days. At 2 days post-confluence, the medium was changed to DMEM/ 10% FBS supplemented with isobutyl-methylxanthine (0.5 mM), dexamethasone (I/zM), bovine insulin (5/~g/ml) and gentamicin (10/zg/ml) for 48 h. The medium was then changed to DMEM/10% FBS supplemented with gentamicin (10/~g/ml) and bovine insulin (5/zg/ml), and the cells were maintained in this medium (replaced every 2 days) until the adipocyte phenotype (rounded morphology and numerous included fat droplets) was evident in most of the cells (5-7 days). Unless stated otherwise, 48 h before 3T3-L1 adipocytes were to be used, the medium was changed to DMEM/10% FBS without the insulin supplement. 2.2. Hormone treatment o f cells and lysate preparation

Pioglitazone (10/zM) was added in DMSO (0.1% v/v, final concentration) and was present for the 24-h period preceding hormone treatment. Cells not exposed to pioglitazone received medium that contained DMSO (0.1%). Except where stated otherwise, neither pioglitazone nor DMSO was included at subsequent steps where the cells were washed then exposed to hormones. After the cells had been pretreated with either pioglitazone or vehicle, they were washed as follows to allow recovery from the effects of insulin and IGF-I (present in the serum used to supplement the culture medium) before the effects of these hormones on PI 3-kinase were examined. DMEM/ 10% FBS was removed from cells which were then washed twice with 1.5 ml of DMEM supplemented with bovine serum albumin (BSA, RIA grade, 10mg/ml) (DMEM/BSA). Cells were then incubated in 1.5 ml of DMEM/BSA for 1 h at 37°C in a CO 2 incubator. This medium was replaced with 0.9 ml of DMEM/BSA/Hepes (20 mM, pH 7.4) and the dishes of cells were placed on a heating block at 37°C. Insulin (porcine) or IGF-I (human recombinant) at various concentrations was added in 0.1 ml of DMEM/BSA/Hepes and, after various periods of exposure, the medium was aspirated and rapidly re-

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

placed with 0.3 ml of ice-cold lysis buffer [Tris-HC1 (20 mM), pH 8.1, NaC1 (137 mM), MgC12 (1 mM), CaC12 (1 mM), glycerol (10% v/v), Nonidet P-40 (1% w/v), sodium orthovanadate (0.15mM), aprotinin (10/zg/ml] (Ruderman et al., 1990). Lysed cells were scraped from dishes and transferred to 1.5-ml microfuge tubes. The remaining cell material was removed from the dishes with a further 0.3 ml of lysis buffer and the combined lysate (0.6 ml) was centrifuged (approx. 10 000 x g for 10 min at 4°C). Supernatant fluid was recovered and stored at -70 ° C until immunoprecipitations were performed.

2.3. lmmunoprecipitation of phosphotyrosyl-proteins Phosphotyrosyl-proteins were immunoprecipitated and PI 3-kinase activity in the immunoprecipitates was measured using modifications of procedures reported previously (Ruderman et al., 1990; Endemann et al., 1990; Serunian et al., 1991). Cell lysates (0.4 ml) were mixed for 1.5 h on a nutator at 4°C with 3/zg of anti-phosphotyrosine (pY) antibody. Protein A-agarose beads (10/~1) suspended in 40/tl of phosphate-buffered saline (PBS) were then added and mixing was continued for 2 h at 4°C. The beads with adherent immunoprecipitates were collected by brief centrifugation (5 s at approx. 5000 x g) and washed sequentially at 4°C as follows: twice with 0.5 ml of PBS containing Nonidet P-40 (1%, w/v) and dithiothreitol (DTT, 1 raM), twice with 0.5 ml of LiC1 (0.5 M) in Tris-HC1 (0.1 M, pH 7.6) with DTT (I mM), and twice with 0.5 ml of NaCI (0.1 M) in Tris-HC1 (10 mM, pH 7.4) with EDTA (1 mM) and DTI" (1 mM) (Ruderman et al., 1990). Traces of the last wash solution were removed using a filter paper wick and PI 3-kinase activity of the immunoprecipitates was assayed immediately. In some experiments, anti-pY antibodies were replaced by anti-PI 3-kinase antibodies (3/zl) and all other operations were identical. 2.4. Assay of Pl 3-kinase activity PI 3-kinase activity was routinely measured directly on pY-immunoprecipitates on the agarose beads using PI and [y-32p]ATP as substrates. Stock suspensions of PI (0.625 mM)(from porcine liver, Serdary Laboratories Inc.) were prepared in assay buffer [Hepes (20 mM, pH 7.1), EGTA (0.4 mM), Na2HPO4 (0.4 mM)] by use of ultrasound (probe sonicator, 50 W for five 15-s periods). Immunoprecipitates were warmed to room temperature (22°C) and 5/zl of BSA (10 mg/ml) and 20/zl of PI were added. After 5 min, 25/zl of [y-32p]ATP (80/zM, approx. 2 Ci/mmol) in assay buffer containing MgC12 (20 mM) were added and the reaction allowed to proceed for 7 min at room temperature. Reactions were terminated with the addition of 0.15 ml of chloroform/methanol/HC1 (1 N) (24:24:6, by vol.). After vigorous mixing and centrifugation (approximately 10 000 x g for 1 min), 30/zl of the lower organic phase were applied to TLC plates (silica gel Type 60, Merck) that had been impregnated with potas-

3

sium oxalate (1%) and activated at ll0°C for 1 h. TLC plates were developed in n-propanol/acetic acid (2 M) (13:7, v/v) for 5 h (Ruderman et al., 1990). Authentic PI, PI 4-P and PI 4,5-P 2 migrated in this TLC system with Rf values of 0.82, 0.60 and 0.34, respectively. PI 3-P was not resolved from PI 4-P in this system. Developed TLC plates were labeled with radioactive ink (to facilitate location of radiolabelled products) and radioactivity incorporated into [32p]PI 3-P was measured using a radiochromatogram scanner (Ambis Systems Inc.) with approximately 40% efficiency. In some experiments, [32p]PI 3-P quantitated using the radiochromatogram scanner was scraped from TLC plates and quantitated again by liquid scintillation counting. Using these two estimates of 32p incorporated into PI 3-P, and the specific radioactivity of the [y-32p]ATP used, pmol of PI 3-P synthesized was computed. Protein in lysates was measured using the BCA assay kit (Pierce Chemical Company) with BSA for calibration. PI 3-kinase activity was expressed as pmol PI 3-P formed min -~ mg lysate protein-1. The amount of PI 3-kinase activity measured was proportional to the amount of lysate used (to at least 1.2 mg protein). Only one radiolabelled product was detected by TLC and had an Rf value (0.6) expected for PI 3-P. Although PI 3-P is not resolved from PI 4-P in this TLC system, [32p]PI 4-P is unlikely to complicate the measurement of [32p]PI 3-P because PI 4-kinase is not immunoprecipitated by pY antibodies and, in the absence of detergent, PI 4-kinase does not catalyze the in vitro phosphorylation of PI (Carpenter and Cantley, 1990). In some experiments, either PI 4-P or PI 4,5-P 2 was substituted for PI as substrate in the PI 3-kinase assay. Stock suspensions of PI 4-P and PI 4,5-P2 were prepared as described for PI and all other aspects of the assay were identical. When PI 4-P and PI 4,5-P2 were tested as in vitro substrates for PI 3-kinase, the rate of 32p incorporation was much less than that with PI as substrate. Maximal PI 3-kinase activity, with either PI 4-P or PI 4,5-P 2 as substrate, was < 10% of that with PI as substrate. Furthermore, at concentrations greater than 0.1 mM, pronounced substrate inhibition was observed with PI 4-P and PI 4,5-P2, but not PI. For these reasons, PI was used as a convenient in vitro substrate for PI 3-kinase in subsequent experiments.

2.5 Measurement of Pl 3-kinase activity in mouse adipocytes Adipocytes were isolated from epididymal fat pads of C57BL/6J mice by collagenase dispersion from tissue minces (Rodbell, 1964). Cells were washed twice with DMEM/BSA/Hepes (20 mM, pH 7.4) and then resuspended in the same medium at approximately 1 ml packed cells per 7 ml of medium. Following treatment with insulin (at various concentrations for 1-60 min), portions (1 ml) of cell suspensions were transferred to 1.5 ml size microfuge tubes that contained 0.35 ml of

4

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

diisononyl phthalate oil. Tubes were centrifuged (5 s at approximately 10 000 x g) then rapidly immersed in liquid N2. The layer of frozen cells was excised and transferred to a tube containing 0.6 ml of lysis buffer (see above). Immunoprecipitation and measurement of PI 3kinase activity were then performed using 0.4 ml of lysate as described above.

2.6. Measurement of cyclic AMP phosphodiesterase activity A membrane fraction of 3T3-L1 adipocytes was prepared as described for rat epididymal fat cells, by Kono et al. (1975), except that homogenization in a Dounce homogenizer was preceded by a 5-s cell disruption with ultrasound (microprobe sonicator, approx. 35 W). Cyclic AMP phosphodiesterase activity was measured as the rate of formation of [3H]adenosine from [3H]cyclic AMP in a coupled reaction with snake venom nucleotidase (Kono et al., 1975). 2.7. Quantitation of cyclic AMP Cyclic A M P in TCA extracts of 3T3-L1 adipocytes was quantitated by use of a radioimmunoassay (Bleasdale et al., 1990). 3. Results

3.1. Characteristics of insulin-dependent changes in p Yimmunoprecipitable PI 3-kinase activity in mouse adipocytes Adipocytes that were isolated from epididymal fat pads of C57BL/6J mice and then exposed to insulin at various concentrations exhibited dose-dependent increases in PI 3-kinase activity that could be immunoprecipitated using antibodies against phosphotyrosine (Fig.

.--.0.9 C-

4~ L)

B.

/'\

c"

E

t

.

0.3

©

m

E O_

0

T///i

\

< Tu~ 0.6 ® E C

3.2. Insulin- and IGF-I-dependent changes in pYimmunoprecipitable PI 3-kinase in 3T3-L1 adipocytes The amount of pY-immunoprecipitable PI 3-kinase activity in unstimulated 3T3-L1 adipocytes was more than 10-fold greater than that in unstimulated adipocytes from C57BL/6J mice and was increased further when 3T3-LI adipocytes were exposed for 5 min to either insulin or IGF-I (Fig. 2). The 5-10-fold increase in immunoprecipitable PI 3-kinase in 3T3-L1 adipocytes induced by insulin was less than the 20-50-fold increase observed in C57BL/6J adipocytes, but the maximal insulin-induced activity was similar in the two cell types (range of 1.23.1 pmol min -1 mg protein -1, 8 experiments). In 3T3-L1 adipocytes the response to insulin was more rapid than that in C57BL/6J adipocytes, reaching a maximum within 2 min of exposure to insulin (Figs. 1, 3). Furthermore, pY-immunoprecipitable PI 3-kinase in 3T3-L1 adipocytes declined between 3 and 10 min after insulin addition then remained constant for the remainder of the period of exposure to insulin (40 min) (Fig. 3). The maximal increase in pY-immunoprecipitable PI 3kinase activity induced by IGF-I in 3T3-L1 adipocytes was only about 30% of that observed with insulin (Fig. 2) It is unlikely that the effect of IGF-I is mediated by insu-

A.

r')

x:

1A). Immunoprecipitable PI 3-kinase activity in unstimulated adipocytes was low (<0.03 pmol min -1 mg protein -I) but was increased 20-50-fold by insulin. Halfmaximal effects of insulin were observed at concentrations of 3-5 nM. Insulin-induced increases in pYimmunoprecipitable PI 3-kinase activity occurred in two phases; a rapid initial phase within 3 min of insulin addition that accounted for about half of the total response to insulin, and a second slower phase that persisted throughout the exposure to insulin (60 min) (Fig. 1B).

/ 0

/

f T/•

/

1 10 Insulin (nU)

0.5

' 100

C

Y t

0

i

i

[

I

I

I

I

I

I

I

I

10 20 30 40 50 60 Time ( m i n )

Fig. 1. Insulin-induced changes in pY-immunoprecipitable PI 3-kinase activity in mouse adipocytes. Adipocytes isolated from epididymal fat pads of C57BL/6J mice were exposed either to insulin at various concentrations for 5 min (A), or to insulin (100 nM) for various periods of time (B). Cells were then collected by floatation through oil, flash frozen, lysed, and PI 3-kinase activity in anti-phosphotyrosineimmunoprecipitates was measured as described in Section 2. Data are from a typical experiment and are either averages of two measurements (A) or means _+SD of three measurements (B).

K.M. Sizer et al. /Mol. Cell. Endocrinol. 103 (1994) 1-12

c"

:_s o

/

c)

< To~ 2 E

/

c-

Y

O

o

E

I o

o_ E

/

o

o_

0

1

/// I

I

0"(3.1

1 Hormone

I

10 (nM)

I

100

Fig. 2. Insulin- and IGF-I-induced changes in pY-immunoprecipitable PI 3-kinase activity in 3T3-L1 adipocytes. 3T3-LI adipocytes were washed and allowed to recover from hormones present in the culture medium for 1 h. Cells were then exposed for 5 min to either insulin (O) or IGF-I (O) at the concentrations indicated. Cell lysates were prepared, pY-immunoprecipitates were isolated and PI 3-kinase activity in immunoprecipitates was measured as described in Section 2. Data are averages of duplicate measurements from a typical experiment.

(basal activity). PI 3-kinase activity was also measured in cells that had been washed, allowed to recover and then exposed for 5 min to insulin (10 nM) (insulin-dependent activity), IGF-I (10 nM) (IGF-I-dependent activity), or insulin (500 nM) (insulin-plus IGF-I-dependent activity). Constitutive PI 3-kinase activity was greatest in rapidly growing 3T3-L1 fibroblasts and in 3T3-L1 adipocytes maintained in medium supplemented with insulin (5 ktg/ ml) (Fig. 4). After cells were washed and allowed to recover, basal PI 3-kinase activity was much less than constitutive activity, except in the 3T3-L1 fibroblasts in mid-log phase of growth where it remained high (Fig. 4). Perhaps because basal PI 3-kinase activity remained high in rapidly growing 3T3-L1 fibroblasts, activity was unresponsive to any of the hormone treatments. At all other stages of development, IGF-I (10nM) increased pYimmunoprecipitable PI 3-kinase activity (Fig. 4). The response to insulin (10 nM) was less than that to IGF-I (10 nM) in 3T3-L1 fibroblasts in either late log phase of growth or at confluence. As expected, 3T3-L1 adipocytes, which express many more insulin receptors than do 3T3L1 fibroblasts, exhibited a greater PI 3-kinase response to insulin relative to IGF-I than was observed in the fibroblasts. Unexpectedly, however, the response of 3T3-L1

lin receptors because at the concentrations of IGF-I and insulin that elicit half-maximal PI 3-kinase responses (310 nM) there is negligible receptor cross-reactivity in these cells (Smith et al., 1988). 3.3. Responses of Pl 3-kinase to insulin and IGF-I during 3T3-L1 cell differentiation When 3T3-L1 fibroblasts adopt the adipocyte phenotype, the cell surface abundance of insulin receptors increases greatly whereas IGF-I receptor abundance changes little (Reed et al., 1977; Smith et al., 1988). Consequently, the responsiveness of PI 3-kinase in 3T3-L1 cells to insulin relative to IGF-I may be expected to increase during differentiation. The effects of insulin and IGF-I on pY-immunoprecipitable PI 3-kinase activity were measured in 3T3-L1 cells at five stages: (i) 3T3-L1 fibroblasts in mid-log phase of growth; (ii) 3T3-L1 fibroblasts in late log phase of growth; (iii) confluent, quiescent 3T3-L1 fibroblasts; (iv) 3T3-L1 adipocytes with down-regulated insulin receptors (maintained in medium with high insulin, 5#g/ml); (v) 3T3-L1 adipocytes with up-regulated insulin receptors (maintained in medium with insulin <30 pM). Cells at each of these stages were subjected to various experimental treatments and PI 3kinase activity in pY-immunoprecipitates was measured. Some cells were lysed immediately after removal of the culture medium and pY-immunoprecipitable PI 3-kinase activity was measured (constitutive activity). Other cells were washed and allowed to recover for 1 h in hormonefree medium before PI 3-kinase activity was measured

5

[Insulin]=

1.2

lOnM

1.0 0.8

>,~"

o

1

0.6 0.4

0

o_ 0.2

SE

Bosol (no

insulin)

:=8 0

[IGF-I] =

1.0

lOnM

~d E I

mo -6 -E D_ ~-

0.8 0.6 0.4 0.2 0

0

Time

5

10

after

15

20

hormone

25

30

,35

40

addition (min)

Fig. 3. Kinetics of insulin- and IGF-I-induced changes in PI 3-kinase activity in 3T3-L1 adipocytes. 3T3-L1 adipocytes in DMEM/10% FBS were pretreated with either pioglitazone (10#M) ( e ) or vehicle (DMSO, 0.1% v/v) (O) for 24 h. Cells were then washed, allowed to recover from hormones in the culture medium, and then incubated in DMEM/BSA/Hepes (20 raM, pH 7.4) supplemented with either insulin (10 nM) (A), or IGF-I (10 nM) (B) for various periods of time. Cell lysates were processed for the measurement of PI 3-kinase activity in pY-immunoprecipiates as described in Section 2. Mean values _+ SEM, n=4.

6

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

~;~ 1.5 < Z T °03 °c

ro

constitutive basal IGF-I (lOnM) Insulin (10nM) Insulin (500nM)

. 1.0 Lc

-6

0

I

n.l M Mid-Log L

Late-Log Fibroblasts

Confluent J

I

low IR

high IR [

Adipocytes

Fig. 4. Insulin- and IGF-I-induced changes in PI 3-kinase activity at various stages of 3T3-L1 cell development. 3T3-L1 cells were examined while

expressing the fibroblast phenotype at three stages: (i) mid-log phase of growth; (ii) late-log phase of growth; and (iii) confluent quiescent, and while expressing the adipocyte phenotype under two conditions, i.e. (iv) with down-regulated insulin receptors (low IR) achieved by maintaining the cells in DMEM/10% FBS supplemented with insulin (5/zg/ml) and (v) with up-regulated insulin receptors (high IR) (insulin in culture medium << 100 pM). PI 3-kinase activity in pY-immunoprecipitates from cell lysates was measured immediately after removal of the culture medium (constitutive activity), after washes and a recovery period (basal activity), and after washes, recovery and a 5 min exposure to either IGF-I (10 nM) or insulin (10 and 500 nM). Data are mean values _+SEM, n = 3. adipocytes to I G F - I was also greater than that of growing 3T3-L1 fibroblasts. This increase is even more pronounced when data are normalized per cell rather than per mg cell protein because total protein per cell more than doubles when 3T3-L1 cells adopt the adipocyte phenotype. When the growing 3T3-L1 fibroblasts became confluent there was a large increase in IGF-I-stimulated PI 3kinase activity (Fig. 4). Interestingly, the responses of PI 3-kinase to insulin and to I G F - I were not additive. Exposure of 3T3-L1 adipocytes to insulin (500 nM) (to activate both insulin and I G F - I receptors) did not increase pY-immunoprecipitable PI 3-kinase activity above that observed when only insulin receptors were activated (Fig. 4). The inability to increase pY-immunoprecipitable PI 3kinase activity above that observed after exposure to in-

sulin (10 nM) may indicate that both insulin and I G F - I utilize a common, limiting pool of PI 3-kinase. If there is a common, limiting pool of PI 3-kinase, it is less than the total cellular PI 3-kinase because the amount of PI 3kinase activity that could be immunoprecipitated from lysates of cells exposed to insulin (10 nM) using pY antibodies was less than half of that which could be immunoprecipitated with antibodies against PI 3-kinase (Table 1). A mixture of pY and PI 3-kinase antibodies did not immunoprecipitate more PI 3-kinase activity than that obtained using PI 3-kinase antibodies alone, indicating that the pY-immunoprecipitable PI 3-kinase is contained within the fraction immunoprecipitated by PI 3-kinase antibodies (Table 1).

3.4. Effects of an insulin-sensitizer on insulin- and IGF-Idependent increases in PI 3-kinase activity Table 1

Immunoprecipitable PI 3-kinase activity in lysates of insulin-treated 3T3-LI adipocytes Precipitating antibody

PI 3-kinase activity precipitated (pmol min-1 mg-1 protein)

Anti-phosphotyrosine Anti-P1 3-kinase Anti-phosphotyrosine plus anti-Pl 3-kinase

0.89 _+0.02 2.08 -+0.01 1.82 _+0.05

Mean values _+SEM, n = 3. Lysates from 3T3-L1 adipocytes that had been exposed for 5 min to insulin (100 nM) were prepared for the immunoprecipitation of PI 3kinase activity using anti-phosphotyrosine antibody (3 fig), anti-PI 3kinase antibody (3/4) or a combination of these antibodies. PI 3-kinase activities are expressed per mg of lysate protein and have not been corrected for any inhibition produced by the precipitating antibodies.

The insulin-sensitizer, pioglitazone, is only sparingly soluble in water and so was added to culture medium as a solution in D M S O (0.1% final concentration). Neither pioglitazone nor D M S O directly affected either the immunoprecipitation or the PI 3-kinase assay (data not shown). 3T3-L1 adipocytes were initially exposed to pioglitazone (10/~M) for 24 h because previously it was found that other effects of pioglitazone in these cells were all maximal by 10/~M but required more than 12 h to develop. Treatment of 3T3-L1 adipocytes with pioglitazone (10ffM) for 24 h, however, did not greatly influence either the magnitude or kinetics of changes in PI 3-kinase activity induced by either insulin (Fig. 3A) or IGF-I (Fig. 3B). Furthermore, the sensitivity of these responses to insulin and IGF-I were not greatly affected by treatment of the 3T3-L1 adipocytes with pioglitazone (10/~M) for

K.M. Sizer et al./Mol. Cell. Endocrinol. 103 (1994) 1-12

B,

A. 1o-%

:>~c: .~ o2

7

"S %

[3_

E

(f) U

.

./7, o/4/

I c

Zo--o

I

I//

I

0 0.1

I

I

I 10 Insulin (nM)

I

I //

100

i

0 0.I

I

i

i

1 10 IGF-I (riM)

100

Fig. 5. Effect of pioglitazone on insulin- and IGF-I-induced changes in PI 3-kinase activity in 3T3-L1 adipocytes. 3T3-L1 adipocytes were treated with either pioglitazone (10/zM) (e) or DMSO (0.1%) (O) in DMEM/10% FBS for 24 h, then washed, allowed to recover from hormones in the culture medium, and exposed to either insulin (A) or IGF-I (B) at various concentrations for 5 min. PI 3-kinase activity in pY-immunoprecipitates was measured as described in Section 2. Mean values _+SEM, n = 4. 24 h (Fig. 5). In other experiments, acute exposure to pioglitazone was likewise found to have negligible effect on the PI 3-kinase response to either insulin or IGF-I (data not shown). 3T3-L1 fibroblasts, which have an IGFI receptor abundance similar to that of 3T3-L1 adipocytes, also exhibited IGF-I dependent changes in PI 3 kinase that were unresponsive to pioglitazone ( 1 0 # M , for 24 h) (data not shown).

changes in PI 3-kinase (Fig. 6). In well-differentiated 3T3-L1 adipocytes, isoproterenol reduced maximal responsiveness by about 35%. Pretreatment with pioglitazone ( 1 0 # M , for 24 h) prevented this attenuation by isoproterenol of insulin-induced changes in PI 3-kinase (Fig.

T

3.5. Attenuation by isoproterenol of insulin-dependent changes in PI 3-kinase and protection by pioglitazone The inability to demonstrate an effect of pioglitazone on insulin-induced changes in PI 3-kinase activity may be because 3T3-L1 adipocytes are already fully responsive to insulin. Pioglitazone is known to dramatically reduce insulin and glucose concentrations in the blood of insulin resistant animals, but has only minor effects in normal animals (Bleasdale and Swanson, 1993). Beta-adrenergic compounds administered to normal rats were found to induce a form of insulin-resistance that is overcome by ciglitazone, a thiazolidinedione analog of pioglitazone (Kirsch et al., 1984). Therefore, we examined the possibility that isoproterenol may antagonize insulin-induced changes in PI 3-kinase, and that pioglitazone may protect against this antagonism. 3T3-L1 adipocytes were exposed to pioglitazone ( 1 0 # M ) or D M S O for 24 h, washed and maintained in hormone-free medium for 1 h. Cells were then stimulated for 15 min with either isoproterenol ( 1 0 # M ) or various cell-penetrating analogs of cyclic A M P (1 mM). Insulin was added for the final 5 min of exposure to the above agonists. Cells were then lysed and pY-immunoprecipitable PI 3-kinase activity was measured. Treatment of 3T3-L1 adipocytes with isoproterenol altered the dose-response curve for insulin-induced

2>

°--

©

< ¢~

4-, O

n

Tc~ E

"e 5

c_

'.~

~

E

I

0

0.1

1

10

100

Insulin (nM) Fig. 6. Effect of pioglitazone on the antagonism by isoproterenol of insulin-induced changes in pY-immunoprecipitable PI 3-kinase activity. 3T3-L1 adipocytes were pretreated for 24 h with either pioglitazone (10#M) or DMSO (0.1%) and then washed and allowed to recover from hormones in the culture medium. Pioglitazone (10#M) (A) or DMSO (0.1%) (O,e) was present in the wash solutions and at all subsequent steps. Cells were exposed for 15 min to either isoproterenol (10/tM) (e,A) or vehicle (O) and for the final 5 min of this exposure period, insulin at various concentrations was also added. PI 3-kinase activity in pY-immunoprecipitates was measured as described in Section 2. O, insulin alone; e, insulin after pretreatment with isoproterenol; A, insulin after pretreatment with pioglitazone then isoproterenol. Mean values _+SD, n = 3.

8

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

1500 T

c

I:Z

o 1ooo

>,~ ©

o

E

w 500

el_

0

2

Time

after

4

6

8

10

i s o p r o t e r e n o l a d d i t i o n (rain)

Fig. 7. Effect of pioglitazone on isoproterenol-induced cyclic AMP accumulation by 3T3-L1 adipocytes. 3T3-LI adipocytes were maintained for 24 h in DMEM/10% FBS supplemented with either pioglitazone (10/zM) (D,II) or vehicle (DMSO, 0.1%) (O,O,A,A), and then washed as described in Section 2. Either pioglitazone or DMSO was present in the wash solutions and at all subsequent steps. The medium was changed to DMEM/BSA/Hepes (20 mM, pH 7.4) supplemented with insulin (5 nM) and either pioglitazone (10/.tM) (EI,N,A,A) or DMSO (0.1%) (O,O) with (closed symbols) or without (open symbols) isobutylmethylxanthine (1 mM). After 5 min, isoproterenol (10,uM) was added, and at various intervals thereafter, cyclic AMP in TCA extracts of the cells was quantitated (see Section 2). O,o, isoproterenol alone; A,A, isoproterenol after 5 min pretreatment with pioglitazone; B,m, isoproterenol after 24 b pretreatment with pioglitazone. Data are mean values _+ SD, n = 3 and are typical of four experiments.

6). Unlike previous experiments, pioglitazone (or DMSO) was present throughout the procedure (including cell washes) in the experiments reported in Figs. 6 and 9. Even when present continuously, however, pioglitazone

>"

5O

:>

t-

O

-,-'

T

<

o

C.O U

'(i,)

~5

l

'" 4 0

(D

\_

I

E

oZ \

= 30

.--

E

o

~

20

EL 0

I

0

I

I

I 10 Pioglitazone (,u,M)

Fig. 8. Acute effect of pioglitazone on insulin-stimulated cyclic AMP phosphodiesterase activity in 3T3-L1 adipocyte membranes. 3T3-L1 adipocytes were incubated for 5 min in DMEM/BSA/Hepes (20 mM, pH 7.4) supplemented with either insulin (10 nM) (O) or vehicle (O). Cyclic AMP phosphodiesterase activity in a membrane fraction prepared from these cells was measured at 30°C (see Section 2) in the presence of pioglitazone added at various concentrations 7 min before the addition of substrate. Mean values _+SD, n = 4.

still did not reproducibly alter either basal or maximally insulin-stimulated PI 3-kinase activity (Figs. 6 and 9). These data indicate that pioglitazone does not have an effect on PI 3-kinase that is rapidly lost when pioglitazone is removed from the cells.

3.6. Effects of pioglitazone on isoproterenol-induced accumulation of cyclic AMP Because cyclic AMP analogs also attenuated insulininduced changes in PI 3-kinase (see below), it was suspected that the protective effect of pioglitazone may involve inhibition of fl-receptor-coupled generation of cyclic AMP. It was found, however, that pretreatment with pioglitazone (10#M for 24 h) did not inhibit isoproterenol-induced accumulation of cyclic AMP, but rather caused a small but significant potentiation (Fig. 7). This potentiation was observed in either the presence or absence of a cyclic AMP phosphodiesterase inhibitor, isobutylmethylxanthine (Fig. 7). Acute exposure to pioglitazone (10#M) had negligible effect on isoproterenolinduced cyclic AMP accumulation (Fig. 7). 3.7. Acute effect of pioglitazone on insulin-stimulated high affinity cyclic AMP phosphodiesterase activity High affinity cyclic AMP phosphodiesterase activity was measured in membranes isolated from 3T3-L1 adipocytes that had been exposed to insulin (10 nM) or vehicle for 5 min. Cyclic AMP phosphodiesterase activity in membranes from insulin-stimulated cells was approximately 50% greater than in unstimulated cells (Fig. 8). Addition of pioglitazone to membranes isolated from

K.M. Sizer et al./Mol. Cell. Endocrinol. 103 (1994) 1-12

.~ 1.0 > --6 -FJ L

.~ 0

[~7

control

I

insulin

9

~ isoproterenol + insulin ~ db-cAMP + insulin 8 - B r cAMP + insulin 8-(4-chlorophenylthio)-c.AMP + insulin

o_


E

I

I-"b -6 --

EL.

E

,-,

I DMSO

I

Pioglitazone

Fig. 9. Antagonism of insulin-dependent changes in pY-immunoprecipitable PI 3-kinase and the effects of pioglitazone, 3T3-L1 adipocytes were pretreated with either pioglitazone (10/~M) or DMSO (0.1%) for 24 h and then washed and allowed to recover from hormones in the culture medium. Pioglitazone (10ffM) or DMSO (0.1%) was present in the wash solution and at all subsequent steps. Cells were then treated for 15 min with either isoproterenol (10/~M) or one of several analogs of cyclic AMP (1 mM). For the last 5 min of exposure to these agonists, insulin (5 nM) was also present. PI 3-kinase activity in pY-immunoprecipitates was measured as described in Section 2. Mean values _+ SEM, n = 3. Abbreviations: db-cAMP, N 6, 2'-O-dibutyryl-adenosine 3':5'-cyclic monophosphate; 8-Br-cAMP, 8-bromoadenosine 3':5'-cyclic monophosphate; 8-(4-chlorophenylthio)-cAMP, 8(4-chlorophenylthio)-adenosine 3':5'-cyclic monophosphate. *Statistically different from the corresponding condition of insulin alone (with DMSO or pioglitazone) at p < 0.05 (two-tailed Student's t-test).

either control or insulin-stimulated cells resulted in a dose-dependent inhibition of cyclic AMP phosphodiesterase activity (Fig. 8).

3.8. Attenuation by cyclic AMP analogs of insulindependent changes in PI 3-kinase and protection by pioglitazone 3T3-L1 adipocytes were pretreated with pioglitazone (10ffM) or vehicle for 24 h, washed and then exposed for 15 min to either isoproterenol (10~M) or cell penetrating analogs of cyclic AMP (1 mM). Insulin (5 nM) was added for the final 5 rain of the incubation. As described above, isoproterenol attenuated insulin-induced changes in PI 3kinase and this attenuation was mimicked by 8-bromo cyclic AMP and 8-(4-chlorophenylthio)-cyclic AMP, but not dibutyryl cyclic AMP (Fig. 9). Attenuation by isoproterenol, 8-bromo cyclic AMP and 8-(4-chlorophenylthio)cyclic AMP was prevented either partially or completely in cells pretreated for 24h with pioglitazone (10ffM) (Fig. 9). 4. Discussion The insulin-resistance of KKAY mice (like that in most cases of NIDDM) is a consequence not of a defect in the insulin receptor but an abnormality in an early event in insulin signal transduction. This defect, while not yet identified, appears to be corrected (or compensated for) by pioglitazone (Bleasdale and Swanson, 1993; Hofmann et al., 1991). Because activation of PI 3-kinase appears to be an early event in insulin signal transduction, we exam-

ined the possibility that PI 3-kinase is a target for pioglitazone action. We found that insulin-induced changes in pY-immunoprecipitable PI 3-kinase, which are observed in adipocytes isolated from normal C57/BL6J mice, are essentially totally absent in adipocytes from insulinresistant KKAY mice, and that treatment with pioglitazone partially restores responsiveness of PI 3-kinase to insulin (Dailey et al., 1993). Investigation of the mechanism of this effect of pioglitazone in KKAY mice is complicated by the increase in insulin receptor abundance in liver (Bleasdale and Swanson, 1993) (and most likely other target tissues) that results from the reduction in hyperinsulinemia caused by pioglitazone (Bleasdale and Swanson, 1993; Hofmann et al., 1991). Therefore, the effects of pioglitazone on insulin-dependent changes in PI 3kinase were examined further using as an experimental model 3T3-L1 adipocytes, which were found previously to respond to pioglitazone (Kletzien et al., 1992). We found that 3T3-L1 adipocytes exposed to insulin exhibit an increase in pY-immunoprecipitable PI 3-kinase activity that was similar to that observed in adipocytes from C57/BL6J mice in both magnitude of the maximal insulin-stimulated activity and the insulin concentration required for a half-maximal response. In 3T3-L1 adipocytes, however, pY-immunoprecipitable PI 3-kinase activity in unstimulated cells is much greater than that in unstimulated mouse adipocytes and the kinetics of insulin-induced changes differ. In 3T3-L1 adipocytes, the maximal response to insulin occurs within the initial 2 min of exposure to the hormone and, as in CHO cells (Ruderman et al., 1990), there is no second phase of insu-

10

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

lin-induced increase in PI 3-kinase. In contrast, approximately half of the response of PI 3-kinase during a 60min exposure of C57BL/6J adipocytes to insulin occurs during the initial 2-3 min and the remainder of the increase occurs between 3 and 60 min of exposure to insulin. While the present investigation was in progress, insulin-dependent changes in PI 3-kinase in rat adipocytes were reported (Kelly et al., 1992, 1993; Giorgetti et al., 1992). As in mouse adipocytes, pY-immunoprecipitable PI 3-kinase activity in insulin-stimulated rat adipocytes is half-maximal at 2 min but does not reach a maximum until >20 min after insulin addition and remains elevated for at least 50 min (Kelly et al., 1992). The sustained increase in pY-immunoprecipitable PI 3-kinase activity appears to be a common response to insulin in a variety of cell types and is accompanied by an accumulation of PI 3kinase products that also is prolonged (Ruderman et al., 1990). 3T3-L1 adipocytes also exhibited IGF-I-induced changes in PI 3-kinase. These changes, while smaller than those induced by insulin, occur at concentrations of IGF-I where IGF-I receptors (but not insulin receptors) are activated (Smith et al., 1988). The abundance of insulin receptors on 3T3-L1 adipocytes is much greater than that of IGF-I receptors (Smith et al., 1988). In contrast, in 3T3LI fibroblasts (on which there are more IGF-I receptors than insulin receptors (Smith et al., 1988)) the response of PI 3-kinase to IGF-I is as great or greater than that to insulin. When 3T3-L1 cells are treated to induce the adipocyte phenotype, the cell surface abundance of insulin receptors and the relative responsiveness of PI 3-kinase to insulin increase. These increases are even greater than those depicted in Fig. 3 when the data are normalized to cell number rather than cell protein because adoption of the adipocyte phenotype is accompanied by a substantial (>100%) increase in cell protein. Surprisingly, when 3T3L1 adipocytes are maintained in medium supplemented with insulin at high concentration and then subsequently challenged with insulin (10 nM), the PI 3-kinase response is not much less than that observed in cells maintained in low insulin medium even though the abundance of cell surface insulin receptors and IRS-1 are decreased approximately 50% and 90%, respectively, by the exposure to high insulin (Ronnett et al., 1982; Rice et al., 1993). When 3T3-L1 adipocytes are exposed to insulin at 500 nM (a concentration at which both insulin and IGF-I receptors are activated), pY-immunoprecipitable PI 3kinase activity is not greater than that observed when insulin (10 nM) is used (where only insulin receptors are activated). The reason that the insulin- and IGF-Idependent changes in PI 3-kinase are not additive does not appear to be a limiting common pool of PI 3-kinase. The amount of PI 3-kinase activity that could be immunoprecipitated using anti-PI 3-kinase antibodies was more than twice that which could be immunoprecipitated from lysates of cells stimulated with 100 nM insulin using

anti-phosphotyrosine antibodies. The possibility that not all of the total cellular PI 3-kinase is available for activation by either insulin or IGF-I cannot be precluded. The amount of PI 3-kinase activity immunoprecipitable from lysates of insulin-treated 3T3-L1 adipocytes was similar when either anti-phosphotyrosine or anti-IRS-1 antibodies were used (unpublished data). Furthermore, after immunoprecipitation with anti-phosphotyrosine antibodies, treatment of the resulting supernatant with anti-PI 3kinase antibodies (but not anti-IRS-1 antibodies) yielded a second immunoprecipitate containing PI 3-kinase activity (unpublished data). Recent evidence supports a mechanism in which the activation of PI 3-kinase by insulin requires association of PI 3-kinase with IRS-1, but not the phosphorylation of PI 3-kinase on tyrosine residues (Backer et al., 1992; Myers et al., 1992; Folli et al., 1992). Computation of the fraction of total PI 3-kinase that undergoes insulin-dependent association with IRS-1 cannot be reliably performed using this data. This is because, even in insulin-stimulated cells, some of the PI 3kinase immunoprecipitated using anti-PI 3-kinase antibodies is not associated with IRS-1 and has a reduced specific activity. Specific activity of this PI 3-kinase is increased 2-3-fi~,ld upon addition of either tyrosinephosphorylated IRS-1 (Backer et al., 1992) or peptide fragments thereof (Carpenter et al., 1993). Even in a single cell type estimates of the amount of PI 3-kinase that remains unassociated with IRS-1, even after maximal stimulation with insulin, vary from 30% (Backer ~t al., 1993) to 70% of the total PI 3-kinase (Backer et al., 1992). These in vitro estimates of PI 3-kinase activation do not include any increased activity that may result from subcellular translocation of PI 3-kinase in vivo (Kelly and Ruderman, 1993). While the present data do not address the state of phosphorylation of PI 3-kinase, they indicate that most or all of the insulin-responsive PI 3-kinase in 3T3-L1 adipocytes becomes associated with IRS-1 very soon after exposure to insulin. When 3T3-L1 adipocytes were exposed for 24 h to pioglitazone, there was negligible effect on the kinetics, magnitude, and responsiveness of insulin-dependent changes in pY-immunoprecipitable PI 3-kinase activity. Pioglitazone was also without effect when it was added acutely to 3T3-L1 adipocytes and when added either acutely or chronically to 3T3-L1 fibroblasts. Similarly, pioglitazone had negligible effect on IGF-I responsive PI 3-kinase activity. Because the responses of 3T3-L1 adipocytes to insulin resemble most closely those in adipocytes from normal mice, and pioglitazone is known to greatly improve insulin-sensitivity in insulin-resistant mice but not in normal mice (Bleasdale and Swanson, 1993; Hofmann et al., 1991), we examined the effect of pioglitazone on insulin-responsive PI 3-kinase in 3T3-L1 adipocytes made 'insulin-resistant' by pretreatment with isoproterenol. In adipocytes isolated from normal rats that had been pretreated with isoproterenol, there was a loss of

K.M. Sizer et a l . / Mol. Cell. Endocrinol. 103 (1994) 1-12

almost half of insulin-stimulated glucose uptake and ciglitazone (an anti-diabetic thiazolidinedione analog of pioglitazone) protected against this loss (Kirsch et al., 1984). We found that exposure to isoproterenol for 10 min before, and during a 5-min exposure to insulin, resulted in attenuation of insulin-dependent changes in PI 3-kinase activity. The effect of isoproterenol on the insulin dose-response curve was very similar to the effect of isoproterenol on insulin-dependent glucose uptake by rat adipocytes reported previously (Kirsch et al., 1984), and the dose-response curve was normalized in those cells which had been exposed for 24 h to pioglitazone. Because it has been proposed that fl-adrenergic attenuation of insulin responsiveness in adipocytes may involve phosphorylation of the fl-subunit of the insulin receptor by a cyclic AMP-dependent protein kinase (H~iring, 1991), we examined the possibility that pioglitazone was blocking isoproterenol-induced generation of cyclic AMP. We found, however, that acute exposure of 3T3-L1 adipocytes to pioglitazone did not affect isoproterenol-induced accumulation of cyclic AMP measured in either the absence or presence of a cyclic AMP phosphodiesterase inhibitor. Furthermore, exposure to pioglitazone for 24 h resulted in a potentiation of isoproterenol-induced cyclic AMP accumulation. Addition of pioglitazone to membranes isolated from 3T3-L1 adipocytes resulted in a dose-dependent inhibition of insulin-responsive high affinity cyclic AMP phosphodiesterase activity. However, this acute effect of pioglitazone is unlikely to be responsible for the potentiation of isoproterenol-induced cyclic AMP accumulation, because acute exposure of 3T3-L1 adipocytes did not alter the cyclic AMP response and because the potentiation was observed in both the absence and presence of a phosphodiesterase inhibitor. Furthermore, because insulin-dependent changes in pYimmunoprecipitable PI 3-kinase were also attenuated by cell-penetrating analogs of cyclic AMP, and pioglitazone also protected against this attenuation, the interference by pioglitazone in isoproterenol-attenuation of insulin responses is most likely distal to fl-adrenergic receptorcoupled production of cyclic AMP. Recently, it was reported that phosphorylation of a serine residue in the 85 kDa subunit of PI 3-kinase results in an inhibition of kinase activity (Carpenter et al., 1993). While the protein kinase that catalyzes this phosphorylation is not a cyclic AMP-dependent protein kinase (Carpenter et al., 1993), the possibility that pioglitazone may alter the phosphorylation state of PI 3-kinase can not be precluded. In summary, both insulin and IGF-I are able to increase pY-immunoprecipitable PI 3-kinase activity in 3T3-L1 adipocytes. The relative responsiveness to insulin increases as the cells adopt the adipocyte phenotype. PI 3kinase responses to insulin and IGF-1 are not additive. Pretreatment with pioglitazone for 24 h does not greatly affect the sensitivity or kinetics of either insulin- or IGFI-dependent changes in PI 3-kinase activity in 3T3-L1

11

adipocytes. However, in adipocytes in which insulin resistance has been induced by pretreatment with isoproterenol, pioglitazone increases the insulin-dependent component of pY-immunoprecipitable PI 3-kinase activity. Thus, while chronic pioglitazone treatment appears to have no direct effect on insulin stimulated PI 3-kinase activity, it may interfere with a mechanism that normally antagonizes this action of insulin. These findings suggest that the facilitation of insulin signal transduction by pioglitazone may be indirect and warrant further investigation.

References Backer, J.M., Myers, M.G. Jr., Shoelson, S.E., Chin, D.J., Sun, X.J., Miralpeix, M., Hu, P., Margolis, B., Skolnik, E.Y., Schlessinger, J. and White, M.F. (1992) EMBO J. 11, 3469-3479. Backer, J.M., Myers, M.G. Jr., Sun, X.J., Chin, D.J., Shoelson, S.E., Miralpeix, M. and White, M.F. (1993) J. Biol. Chem. 268, 82048212. Bleasdale, J.E. and Swanson, M.L. (1993) Biochim. Biophys. Acta 1181,240-248. Bleasdale, J.E., Thakur, N.R., Gremban, R.S., Bundy, G.L., Fitzpatrick, F.A., Smith, R.J. and Bunting, S. (1990) J. Pharmaeol. Exp. Ther. 255,756-768. Carpenter, C.L. and Cantley, L.C. (1990) Biochemistry 29, 1114711155. Carpenter, C.L., Auger, K., Duckworth, B.C. Hou, W.-M., Sehaffhausen, B. and Cantley, L.C. (1993) Mol. Cell. Biol. 13, 16571665. Colca, J.R. and Morton, D.R. (1990) in New Antidiabetic Drugs (Bailey, C.J. and Flatt, P.R., eds.), pp. 255-261, Smith-Gordon, New York. Dailey, C.F., Palazuk, B.J., Jacob, C.S., Bleasdale, J.E. and Colca, J.R. (1993) Diabetes 42, 116A. Endemann, G., Yonezawa, K. and Roth, R.A. (1990) J. Biol. Chem. 265, 396-400. Folli, F,, Saad, M.J.A., Backer, J.M. and Kahn, C.R. (1992) J. Biol. Chem. 267, 22171-22177. Giorgetti, S., Balloni, R., Kowalski-Chauvel, A., Cormont, M. and van Obberghen, E. (1992) Eur. J. Biochem. 207,599-606. Green, H. and Kehinde, O. (1975) Cell 5, 19-27. Htiring, H.U. (1991) Diabetologia 34, 848-861. Hayashi, H., Kamohara, S., Nishioka, Y., Kanai, F., Miyake, N., Fukui, Y., Shibasaki, F., Takenawa, T. and Ebina, Y. (1992) J. Biol. Chem. 267, 22575-22580. Hofmann, C.A., Lorenz, K. and Colca, J.R. (1991) Endocrinology 129, 1915-1925. Kelly, K.L. and Ruderman, N.B. (1993) J. Biol. Chem. 268, 43914398. Kelly, K.L., Ruderman, N.B. and Chen, K.S. (1992) J. Biol. Chem. 267, 3423-3428. Kirsch, D.M., Bachmann, W. and Hfiring, H.U. (1984) FEBS Lett. 176, 49-54. Kletzien, R.F., Foellmi, L.A., Harris, P.K.W., Wyse, B.M. and Clarke, S.D. (I 992) Mol. Pharmacol. 42, 558-562. Kono, T., Robinson, F.W. and Sarver, J.A. (1975) J. Biol. Chem. 250, 7826-7835. Myers, MG. and White, M.F. (1993) Diabetes 42, 643-650. Myers, M.G. Jr., Backer, J.M., Sun, X.J., Shoelson, S., Hu, P., Schlessinger, J., Yoakim, M., Schaffhausen, B. and White, MF. (1992) Proc. Natl. Acad. Sci. USA 89, 10350-10354. Nakanishi, H., Brewer, K. and Exton, J.H. (1993) J. Biol. Chem. 268, 13-16.

12

K.M. Sizer et al. / Mol. Cell. Endocrinol. 103 (1994) 1-12

Olefsky, J.M., Garvey, W.T., Henry, R.R., Billion, D., Matthaei, S. and Freidenberg, G.R. (1988) Am. J. Med. 85, Suppl. 5A, 86-105. Reed, B.C., Kaufmann, S.H., Mackall, J.C., Student, A.K. and Lane, M.D. (1977) Proc. Natl. Acad. Sci. USA 74, 4876-4880. Rice, K.M., Turnbow, M.A. and Garner, C.W. (1993) Biochem. Biophys. Res. Commun. 190, 961-967. Rodbell, M. (1964) J. Biol. Chem. 239, 375-380. Ronnett, G.V., Knutson, V.P. and Lane, M.D. (1982) J. Biol. Chem. 257, 4285-4291. Ruderman, NB., Kapeller, R., White, M.F. and Cantley, L.C. (1990) Proc. Natl. Acad. Sci. USA 87, 1411-1415.

Serunian, L.A., Auger, K.R. and Cantley, L. (1991) Methods Enzymol. 198, 78-87. Smith, P.J., Wyse, L.S., Berkowitz, R., Lan, C. and Rubin, C.S. (1988) J. Biol. Chem. 263, 9402-9408. Stevenson, R.W., Hutson, N.J., Krupp, M.N., Volkmann, R.A., Holland, G.F., Eggler, J.F., Clark, D.A., McPherson, R.K., Hall, K.L., Danbury, B.H., Gibbs, E.M. and Kreutter, D.K. (1990) Diabetes 39, 1218-1227. Sun, X.J., Miralpeix, M., Myers, M.G. Jr., Glasheen, E.M., Backer, J.M., Kahn, C.R. and White, M.F. (1992) J. Biol. Chem. 267, 22662-22672.